Security in Enterprise AI

7. Using LLMs

7. Using LLMs

Learn how to configure and manage LLMs in StackAI—from provider selection to advanced memory and citation settings.

LLMs (Large Language Models) are the core engine behind StackAI agents. In this tutorial, you’ll learn how to select providers, configure system prompts and inputs, and activate advanced features like memory, citations, and fallback logic. Whether you use OpenAI, Anthropic, local models, or your own API keys, StackAI gives you the flexibility to build AI apps that are secure, compliant, and tailored to your organization’s needs.

Summary

  • Choose from major LLM providers (OpenAI, Anthropic, Google, etc.) or connect custom/local models

  • Drag and drop model nodes and switch models with one click

  • Use StackAI’s enterprise keys or connect your own API keys for full control

  • Configure two key fields:


    • Instructions (system prompt) – defines the model’s role and behavior

    • Prompt – combines user inputs, files, and knowledge bases


  • Use the expanded editor for easier prompt writing

  • Connect knowledge bases and private data sources directly into prompts

  • Enable memory settings to control how the model remembers prior interactions

  • Turn on citations to show which docs or files the model referenced

  • Set up guardrails like PII filtering, fallback models, and automatic retries

  • Chain multiple LLMs for specialized steps within one workflow

  • Manage which models appear in the sidebar via feature access control

  • Assign default API keys per provider for centralized billing and usage

  • Build workflows with the right model for every task, securely and efficiently

  • Next up: explore chaining tools and models for complex agent behavior